[Basic Tutorial] Going through basics on Stable Diffusion WebUI (Patreon)

Hi!
I see a lot of you have a sparkling interest on how to make your own AI art on Stable Diffusion. So let's get some basics out of the way, right?
For starters I'll go through some topics with you and you can ask questions in comments because I'm not the best tutorial maker. So I'm sorry in advance if this is not clear enough.

Summary:

1. Checking system requirements
2. Follow a basic install guide
3. Executing the webui-user.bat
4. Configuration
5. Downloading models
6. Testing
7. Highres. fix

------

1. Checking system requirements:

You should have a GPU with at least 6GB VRAM. (8GB Recommended)

Windows 8, 10, 11, macOS and Linux. (I have Windows 11)

A lot of disk space. 10-15GB is minimum, but it will grow fast. (I'm using more than 400gb)

2. Follow a basic install guide:
AUTOMATIC1111 repository installation and running guide 

Stable Diffusion Art - How to install Stable Diffusion on Windows (AUTOMATIC1111) 

Follow this to a point where you can open the webui on your browser.

If you have 8GB RAM I suggest you right click your webui-user.bat and edit with a notepad. Change the line:

set COMMANDLINE_ARGS=

To: set COMMANDLINE_ARGS=--medvram

3. Execute the webui-user.bat and insert the IP adress it shows on console to your browser. It usually is: 127.0.0.1:7860/

(If you need to close your webui, hit ctrl+c on your CMD window and Y for yes when it asks to finish in blocks) 

4. Now let's configure your webui before going anywhere. You don't have a Stable Diffusion model probably. If you do, don't worry about it yet.

Go to your Settings tab on your webui.

On the left panel click on Stable Diffusion option.

Search for Clip skip and change the value in there to 2.

Hit apply settings on top of the page.

On the left panel click on Sampler parameters option.

Change eta (noise multiplier) for DDIM slider to 0.67.

Also change Eta noise seed delta to 31337.

Hit apply settings on top of the page.

On the left panel click on User interface option.

Search for Quicksettings list.

Add this option by typing it after the sd_model_checkpoint which should be there already (Add it if not): sd_vae.

It should look something like "sd_model_checkpoint, sd_vae" or two bubbles with those options on them.

Hit apply settings on top of the page.

Hit Reload UI just in case.

We are done configuring the basic settings.

5. Download a Checkpoint model from Civitai 

Later I'm going to use an example with this one: AbyssOrangeMix2 - Hardcore (It's NSFW).

Place the models you downloaded under your (\stable-diffusion-webui\models\Stable-diffusion) folder.

You should also download its VAE and place it under (\stable-diffusion-webui\models\VAE) folder.

6. Let's test if it's working.

Hit refresh button next to Stable Diffusion checkpoint on top of your screen and select your downloaded checkpoint. Also do the same for SD VAE and select orangemix.pt if you're following me here.

If things are working strangely try shutting down and re-opening your webui again by following what I said in step 3.

Your options should be something like this, with less tabs and options probably.

Insert some simple prompts in positive and negative (the two big fields in the screen). Example:

POSITIVE: (masterpiece), (best quality), 1girl, dress, green hair

NEGATIVE: low quality, worst quality, bad anatomy

Hit Generate button.

It will look different than mine, but if it generated a girl +- like this so you are on the right track and you are ready to start delving more into Stable Diffusion and AI generated images.

7. Highres. fix

Let's try something before wrapping this up. Check the Hires. fix option.
Change the Upscaler to R-ESRGAN 4x+Anime6B.
Hires steps to 15.
Denoising strength to 0.35
(Optional) Upscale by to 1.5 if you're running out of VRAM after doing all steps
Hit this recicle button to use the same seed as before if you liked your result.

Hit generate button.

You can see the details are way better now (I hope lol).

Okay, time to wrap this up. There is a ton of things to understand and learn. Don't worry about what means what in every step we took here. It's actually a long process. Let's go on baby-steps. If you reach this far you are almost walking by yourself.

Next part (separate post) of this tutorial will cover how to make a fanart NSFW scene using many other embeddings.